Following the conclusion of the 2024 High End Munich Audio Show I took a quick 28 hour trip to Thessaloniki, Greece to spend time wetting my fingers in machine oil and engaging in extensive conversation with a master machinist who holds a PhD in Sound Synthesis and a Master’s Degree in Sonic Arts and Media Production. The topic of discussion over one very intensive day: the mechanical principles required to faithfully extract music from record groove walls.
Why would I go through the hassle and expense for such a thing?!
Let’s back up and provide some important history that culminated in this visit to J.I. Agnew of Agnew Analog, maker of cutting lathes, cutterheads and other precision engineered products on the analog production side.
In February 2014 Michael Fremer wrote an article on a hitherto neglected analog setup parameter that would be given the unfortunate name of “zenith error”, a rotation of the stylus about a vertical axis. Writing about the research performed by Wally Malewicz (founder of WAM Engineering and my mentor), Fremer’s article made it quite obvious that aligning the cantilever is – and had only ever been – simply a proxy for what actually mattered: the two groove-wall contact points on the stylus itself. Until then, we had always *unknowingly assumed* that a line drawn between the two points would be perpendicular to the vertical plane through the cantilever. Not so!
Wally passed away in 2018 and in early 2020, Wally’s son, Andrzej Malewicz, and I picked the company back up to continue Wally’s work. Knowing that stylus zenith error had not yet been fully studied, WAM Engineering purchased a Leica laboratory microscope with some very special features. We contracted a materials scientist and got to work figuring out how to reliably and accurately measure stylus zenith error. It proved to be quite difficult given that the physical feature to be measured is transparent, is in the center of other transparent features and needs to be measured against something WAY, WAY out of the focal plane.
The mechanical, electrical and sonic impact of stylus zenith error had not been studied in the journals over the decades. This was a challenge we decided to dive into with the help of Jeff Falk, a mechanical engineer with expertise in finite element analysis and Fred Stanke, a Stanford University EE PhD who spent his whole career in optics, sound propagation and computational analysis.
We started our analysis using computational modeling and revealed some very interesting characteristics. We then began subjective tests to answer two questions: could we hear the effect of zenith error and, if so, to what level? The answers would be a definite “YES!” and multiple people have reported hearing as little as 0.5° of error!
Next, we needed to prove that the electrical characteristics of zenith error would agree with our expectations. We acquired a 1 foot tall stack of various laboratory test records to help in this matter, many of them now quite “unobtanium” and highly respected by recording engineers.
To our dismay, what we expected to find when analyzing zenith error electrically (by playing the test record and analyzing the distortion characteristics of the output) was often NOT what we ended up seeing. This confused and frustrated us for some time so we decided to look closely at the grooves under the microscope. That is where the mystery began to reveal itself.
We found frequent cutting errors on even the most vaunted test records. Cutting zenith errors as much as 5° weren’t tough to find. This kind of mechanical cutting error is fatal if we are expecting to trust a record to help us know our distortion at time of playback!
We then asked a very well known recording engineer to send us a few of his used cutting styli. Cutting styli are designed to have each cutting edge be 45° from the shank centerline. To our amazement, the very first one we measured was 43°/47°. This cutting stylus guarantees crosstalk cut right into the grooves!
Since not a single test record in our collection articulated in their notes the degrees of cutting error in any one of the five geometric parameters we needed to be concerned with, we knew there would not be a single test record that was worthy of our trust for our measurements nor would any of them be useful as a reliable tool for use in playback optimization. (* see footnote) This brought our zenith error research to a sudden halt.
It was time to get to “the source” and visit a number of cutting engineers. We figured if we could look at how lathes and cutting heads are designed we could employ an in-situ measurement method that could offer a degree of certainty of what the cutting angles are at the time of cutting. We met with seven different engineers all of whom use a variety of cutting lathes and cutterhead brands. It soon became clear to us that the very few tools the engineers had at their disposal to ensure proper alignment of the cutting stylus were quite coarse and the methods available to them to KNOW their vertical modulation and cutting rake angles (before or after lacquer spring back) were limited to non-existent. A few of them asked for our help to design measurement tools for them so they COULD know these angles and thereby make more consistent cuts.
One refrain we heard again and again from recording engineers is that much of the knowledge and best practices surrounding cutting lacquers had been partially or wholly lost to the ages. Many of them were having to go through some of the same mistakes that engineers long ago had solved for themselves as these erstwhile mentor-forebears were no longer around to save them the pain and process.
These visits to mastering labs brought us to the conclusion that being able to make a test record with a KNOWN set of cutting angles and measurement error was not going to be easy as the tools needed to provide the level of necessary control simply do not exist. Between the lack of available schematics for some older cutterheads and the virtual non-existence of any shop tools designed to measure in-situ cutting angles, we were in a place where we knew we needed to design our own measurement tools for these cutterheads. OR, if there were a company that was in the process of making lathes and cutterheads perhaps they had the answer?
We were pleased to find two new companies making professional cutting lathes and cutterheads: Sillitoe in Australia and Agnew Analog in Greece. Through our new connections in recording engineering, we were able to make contact with J.I. Agnew. Many email threads and a zoom call later and we had a plan hatched to visit the Agnew Analog Precision Engineering Laboratory in Thessaloniki, Greece. It was our hope that at Agnew Analog we would find the degree of precision necessary to cut the first test record worthy of the name “reference test record.”
On the closing day of the High End Munich show, I took a late night flight to Thessaloniki. The day that would follow would turn to be quite important to our work on vinyl playback optimization. What we learned about tolerances in the cutting discipline cemented our concerns about the relative lack of cutting control. While cutting engineers generally do a nice job of controlling things like pitch speed, groove depth, surface area optimization, modulation control and such, they tend to be poorly equipped to know their cutting symmetry and cutting angles. Since vinyl playback is a mechanical transcription process, these questions are certainly important if we care to extract the maximum level of content faithfully from the groove walls.
J.I. Agnew has created a cutterhead design that removes much of the mystery and offers a level of geometric cutting control that we have not yet seen. His commitment to an exceedingly high level of precision machining is remarkable and inspiring. Using very “old-school” machining techniques, J.I. can often achieve levels of precision not available with CNC machining. His methods are meticulous and certainly time consuming but it was obvious he understood that attending to precision at the sub-micron level is important for a playback medium where audible groove content is easily measured in the nanometer range.
I believe we talked a full hour about the concept of “flatness” alone and how he believes he can achieve it to a known level of measurement error with his shaper built nearly a century ago. Fascinating stuff for a geek like me!
We look forward to learning more from and working with J.I. Agnew. Doing so will help us not only get records cut with greater precision but will also help the rest of us optimize playback of our beloved records.
As a result of this trip, we are better equipped to continue our research into zenith error. Our research team at WAM Engineering are looking forward to its conclusion since there are more parameters that impact our playback experience but have yet to be studied at all or, at very least, merit a re-evaluation as our ability to measure and analyze materials and mechanical relationships is far more advanced than in the heyday of vinyl playback research of the 1950’s through the 1970’s.
Between the opportunities to tighten up control on the cutting side AND on the playback side we are supremely confident that vinyl is only going to sound better and better into the future.
Thank you so very much to my new friends, J.I. and Kate Agnew! I look forward to our next meeting.
*If records have been cut with such a lack of geometric control over the decades, what is the point of trying to improve our playback setup accuracy? In a word: STATISTICS! The degree to which a cut can err on any one of the relevant cutting angles is equally likely to be done in a clockwise manner versus counterclockwise manner. Since all engineers are aiming for the same setup angles (or, at least, a tight range with respect to vertical modulation angle and cutting rake), if we were to take a data set of all errors and plot them on , say, a -10° to +10° horizontal axis, the bulk of the data would gather around the mean point and the entire data set would approximate a normal distribution curve. Therefore, for us to get the highest probability of getting the most out of all records, we should AIM FOR THE MEAN of that distribution curve. That is exactly what we do here at WAM Engineering: determine what the alignment goals were at time of cutting (net of things like lacquer spring back, DMM pre-distortion, etc.) and then AIM for those same goals at playback.
@J.R. -- Did you get around to asking J.I. why he doesn't try his hand at building "a different" transducer -- namely a "phono-cartridge?" If he can build such an "accurate" cutter-head, it seems like he would have the perfect skill-set, tools and knowledge to make a killer phono-cartridge. I find his discussions on absolute phase on LP's [PS audio blog] very insightful -- I ended up purchasing three of those test discs. Likely a very interesting visit.
This is going to blaze trails to better sound for everyone. Vinyl production and reproduction is so amazing!
Large format photographers nearly always seem to sport a beard! I'm guilty of it too, and I no longer do 10x8 photography, but still have a couple of 4x5 cameras. I'm pretty sure there is a correlation between film photography and analogue playback too!